The Performance of African Protected Areas for Lions and Their Prey

Biological Conservation 209 (2017) 137–149

Contents lists available at ScienceDirect

Biological Conservation

journal homepage: www.elsevier.com/locate/bioc

The performance of African protected areas for lions and their prey

P.A. Lindsey a,b,⁎,1,L.S.Petraccaa,c,1,P.J.Funstona,H.Bauerd,A.Dickmand, K. Everatt e, M. Flyman f, P. Henschel a, A.E. Hinks d,S.Kasikig,A.Loveridged, D.W. Macdonald d, R. Mandisodza h,W.Mgoolai, S.M. Miller j,k, S. Nazerali l,L.Siegem,K.Uisebn, L.T.B. Hunter a a Panthera, New York, NY, United States b Mammal Research Institute, Department of Zoology and Entomology, University of Pretoria, South Africa c State University of New York College of Environmental Science and Forestry, Syracuse, NY, United States d Wildlife Conservation Research Unit, The Recanati-Kaplan Centre, Department of Zoology, University of Oxford, United Kingdom e Centre for African Conservation Ecology, Nelson Mandela University, South Africa f Department of Wildlife and National Parks, P. O. Box 131, Gaborone, Botswana g Kenya Wildlife Service, P.O. Box 40241, 00100 Nairobi, Kenya h Zimbabwe Parks and Wildlife Management Authority, PO Box CY140, Causeway, Harare, Zimbabwe i Department of National Parks and Wildlife, P.O. Box 30131, Lilongwe, Malawi j Molecular Ecology and Evolution Programme, Department of Genetics, University of Pretoria, South Africa k Veterinary Genetics Laboratory, Faculty of Veterinary Sciences, University of Pretoria, South Africa l Independent Researcher, Maputo, Mozambique m GIZ-IS, UNDP/GEF, PO Box 28127, Addis Ababa, Ethiopia n Ministry of Environment and Tourism, Private Bag 13306, Windhoek, Namibia article info abstract

Article history: Using surveys of experts associated with 186 sites across 24 countries, we assessed the effectiveness of African Received 11 October 2016 protected areas (PAs) at conserving lions and their prey, identified factors that influence conservation effective- Received in revised form 8 January 2017 ness, and identified patterns in the severity of various threats. Less than one third of sampled PAs conserve lions Accepted 16 January 2017 at ≥50% of their estimated carrying capacity (K), and less than half conserve lion prey species at ≥50% of K. Given Available online xxxx adequate management, PAs could theoretically support up to 4× the total extant population of wild African lions (~83,000), providing a measurable benchmark for future conservation efforts. The performance of PAs shows Keywords: fi Management marked geographic variation, and in several countries there is a need for a signi cant elevation in conservation Budget effort. Bushmeat poaching was identified as the most serious threat to both lions and to wildlife in general. Bushmeat The severity of threats to wildlife in PAs and the performance of prey populations were best predicted by geo- Encroachment graphic-socioeconomic variables related to the size of PAs, whether people were settled within PAs, human/live- Fencing stock densities in neighbouring areas and national economic indicators. However, conservation outcomes for Panthera leo lions were best explained by management variables. PAs tended to be more effective for conserving lions and/ Threats or their prey where management budgets were higher, where photographic tourism was the primary land use, and, for prey, where fencing was present. Lions and prey fared less well relative to their estimated potential carrying capacities in poorer countries, where people were settled within PAs and where PAs were used for neither photographic tourism nor trophy hunting. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction (Juffe-Bignoli et al., 2014), yet some African countries have set aside vast PA networks. For example, Botswana has gazetted 40% of its terres- Protected areas (PAs) are critical to the protection of biodiversity trial area as PAs, Zambia 38% and Tanzania 32% (www.protectedplanet. and habitat integrity (Geldmann et al., 2013). Approximately 209,000 net, accessed October 2016, Botswana Department of Wildlife and Na- PAs exist globally, covering ~15.4% of the world's land and inland waters tional Parks unpublished data). African countries are also home to (Juffe-Bignoli et al., 2014). State-owned terrestrial PAs in Africa cover some of the largest individual PAs. For example, Tanzania's Selous 14.7% of the continent's land area, slightly less than the global average Game Reserve and adjacent buffer zones cover ~90 000 km2,the Luengue-Luiana-Mavinga complex of parks in Angola ~84,200 km2, and Kafue National Park complex in Zambia N 66,000 km2. Furthermore, ⁎ Corresponding author. E-mail address: [email protected] (P.A. Lindsey). several (mainly southern) African countries have established treaties to 1 Authors contributed equally to this work. conserve even larger areas through the establishment of transfrontier

http://dx.doi.org/10.1016/j.biocon.2017.01.011 0006-3207/© 2017 Elsevier Ltd. All rights reserved. 138 P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 conservation areas (TFCAs) (MacKinnon et al., 2015), such as the ~ resources by protected area authorities to mitigate these threats 520,000 km2 Kavango-Zambezi TFCA. (MacKinnon et al., 2015). There have been some attempts to under- PAs contain essential habitat for many of Africa's most iconic, threat- stand the determinants of conservation success for lions in West Africa ened and endemic species (Bergl et al., 2007). Parks and reserves are a (Henschel et al., 2016), and a narrow focus on the role of management central component of sub-Saharan Africa's tourism industry, which cre- interventions such as fencing in influencing conservation outcomes ates millions of jobs and which has been valued at US$25 billion (WTTC, (Packer et al., 2013; Creel et al., 2013). However, little is known about 2016). PAs are thus of key importance from ecological, economic and the performance of individual PAs continent-wide, patterns in the social perspectives, and through the provision and maintenance of eco- threats facing them and the factors that influence their effectiveness logical services (MacKinnon et al., 2015; Van Zyl, 2015). However, while (Geldmann et al., 2013). frequently valuable on national levels, PAs rarely cover their costs at a Thus, we build upon previous work by looking more broadly at the site level (MacKinnon et al., 2015), and can impose significant costs on role of PAs in conservation success, using the African lion as our focal local people through human-wildlife conflict and foregone opportuni- species. We sought to understand, at a protected area level, (1) which ties for using the land for alternatives (Brockington and Igoe, 2006). PAs are currently sustaining lion populations at 50% or above estimated Such issues can undermine political support for government expendi- carrying capacity, (2) what factors are associated with positive conser- ture on PAs and local support for their existence. vation outcomes for lions and their prey, and (3) to understand patterns The importance of PAs to conservation efforts will increase with time in severity of five main threats to African wildlife in PAs, namely: illegal as human populations grow and habitat in unprotected lands is convert- hunting for bushmeat, encroachment by humans for settlement or agri- ed for agriculture and settlement or to compensate for decreased pro- culture, encroachment by livestock for grazing, human-wildlife conflict, ductivity on over-utilised land (Caro, 2015). This is of particular and the poaching of wildlife for non-meat body parts (e.g. ivory, skins, significance in Africa, where the human population is projected to scales, teeth or other products). grow from 1.1 to 2.8 billion by 2060 (Canning et al., 2015). Even under current human population densities, the effectiveness of many 2. Methods PAs at conserving biodiversity is questionable, and many are underperforming (Craigie et al., 2010; Lindsey et al., 2014). This is particularly 2.1. PAs in lion range evident in West and Central Africa (Bouché et al., 2010; Henschel et al., 2014a; Henschel et al., 2015; Bauer et al., 2015a). We assessed the number and area of PAs in lion range, and estimated Human pressures on PAs take various forms, including poaching, en- the potential lion population that could be conserved on such an area. croachment by humans and livestock, mining and deforestation (Okello The potential carrying capacity for lions for each site was estimated and Kiringe, 2004; Lindsey et al., 2014). These anthropogenic pressures using a model that predicts the variation in lion density based on soil on PAs are becoming more severe, yet resources available for manage- type and rainfall (Loveridge, 2009). For the purposes of estimating the ment and protection are often far from adequate (James et al., 1999; area of land under protection in lion range, and estimating potential Mansourian and Dudley, 2008; Lindsey et al., 2016; Henschel et al., lion numbers if those areas were managed optimally, we defined PAs 2016) and there is little information on the impacts of these threats as being state-owned land officially gazetted as a protected area, and on conservation outcomes. In addition, the functionality of PAs is often where wildlife conservation/utilisation is considered to be the primary undermined further by mismanagement and corruption (Smith et al., land use (excluding private land and community ‘conservancies’, 2003). which typically occur on land with customary tenure/ownership). We excluded wildlife areas on private and community land to provide a 1.1. Protected areas and African lion conservation conservative estimate of the lion range that is protected because the legal protection status of such land is variable. However, we do ac- The African lion (Panthera leo) is an iconic and charismatic species knowledge that private and community conservation areas are of high that is highly valued by society (Macdonald et al., 2015). Lions play a conservation value. Our definition included hunting areas and other key ecological role due to their status as apex predators (Ripple et al., local protected designations as well as national parks. We excluded PA 2014), and have significant economic value as drawcards for photo- complexes (individual PAs or groups of contiguous PAs) of graphic tourism and trophy hunting (Lindsey et al., 2007; Lindsey et b1000 km2, except in South Africa, where fencing and intensive man- al., 2012a). The species has significant cultural value to some societies agement allows for the maintenance of lion populations in smaller (in Africa and elsewhere), such as being symbols of royalty, acting as areas (Packer et al., 2013). Consequently, in South Africa, where PAs sports emblems, or being totems. Lions also confer value in some places are fenced, our cutoff for inclusion was 500 km2. A cutoff of 500 km2 through the illegal and legal trade in lion body parts (Williams et al., allowed for the inclusion of some South African reserves, while exclud- 2016). ing very small reserves where management is likely to be so intensive as Despite their social, ecological, and economic value, lions have un- to preclude meaningful comparison with PAs in other parts of Africa. dergone significant declines in numbers and geographic range in recent years. Lion numbers declined ~43% during 1993–2014, with particularly 2.2. Surveys marked declines in West and Central Africa (Bauer et al., 2015a). As few as 23,000 individuals persist in the wild and the species is listed as Vul- We conducted an online questionnaire survey of individuals with nerable on the IUCN Red List (Henschel et al., 2015); in West Africa, they expertise of PAs within lion range (Appendix 1). The survey was de- are considered Critically Endangered (Henschel et al., 2015; Bauer et al., signed to obtain insights into the performance of populations of lions 2015c). Approximately ~56% of lion range has protected area status; and their prey, to understand the main threats to both, and to provide when well managed, these PAs can frequently support high lion densi- insights into the determinants of conservation success. In order to ob- ties (Riggio et al., 2012). tain a larger sample for the surveys, we expanded our definition of Key threats to lions include human-lion conflict, habitat destruction, PAs to include legally recognised conservancies or other wildlife areas depletion of prey populations, targeted poaching of lions for their body occurring on private and community lands. ‘Experts’ were defined as parts and poorly regulated trophy hunting (Bauer et al., 2015a). Howev- those who are working in the PA in the context of management (n = er, the relative importance of those threats in specific PAs is poorly un- 102) or research related to lions or their prey (n = 32). Respondents derstood. Threats to lions and other wildlife are often exacerbated by had a mean of 9.31 ± 1.1 years of experience in the area in question unfavourable policies, political and economic instability and institution- (range 1–40 years) and were identifiedthroughprofessionalnetworks al weakness on the part of state wildlife authorities and lack of adequate and via ‘snowballing’ sampling technique (Atkinson and Flint, 2001). P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 139

We managed to obtain responses from experts from 21 of the 25 known (where 1 was no threat and 5 was a severe threat) (Appendix 3). lion range countries (Bauer et al., 2015a). In addition, data were collect- These questions were congruent with current insights into conservation ed from experts in Ghana, where the presence of lions is questionable, threats from the literature (Appendix 3). ‘Total threat’ scores were cal- and Rwanda, where lions have recently been reintroduced. In total, ex- culated from the sum of the scores allocated to each threat. A manage- perts provided data for 186 PAs. ment capacity score was derived from questions related to various The effectiveness of PAs for conserving lions was calculated by using aspects of management capacity and resource availability scored on a estimates of the population of lions relative to the estimated carrying 1–5 scale of adequacy (Appendix 1). capacity. Lion population estimates were derived from three sources: (1) The literature, where available (24.2% of PAs, accounting for 2.3. Protected areas with survey data ~61.6% of the total lion population for which we could find estimates) (Tumenta et al., 2010; Kiffner et al., 2012; Becker et al., 2012; Cozzi et We used ArcMap v.10.3.1 (ESRI, 2016) to compile geospatial data for al., 2013; Ferreira et al., 2013; Olléova and Dogringar, 2013; Midlane, 175 of the 186 PAs within lion range for which we had survey re- 2014; Rosenblatt et al., 2014; Yirga et al., 2014; Omoya et al., 2014; sponses; boundary data for the remaining 11 protected areas were un- Everatt et al., 2014a; Henschel et al., 2014a; Henschel et al., 2014b; available. Boundaries were derived from the World Database of Henschel et al., 2015; Bauer et al., 2015a; Bauer et al., 2015b; Bauer et Protected Areas (IUCN/UNEP WCMC, 2007). We computed various met- al., 2015c; Bauer and Ryskay, 2016; Bauer et al., 2016). Of these, 56.3% rics for (1) individual PAs and (2) complexes of PAs, the latter of which used call-ups, 43.8% spoor counts, 25.0% individual recognition, and were established by aggregating PAs separated by a maximum of 500 12.5% camera trapping, all of which are considered to be adequately sci- metres. This distance was chosen due to minor inaccuracies in the entific for our purposes (Henschel et al., 2014b; Midlane et al., 2015; boundary data. Accounting for this error allowed adjacent PAs to be Bauer et al., 2015c). (2) From estimates derived from on-going unpub- properly assigned to the same complex. lished research provided by the management authority or scientists working on site (45.0% of PAs accounting for 34.1% of the total lion pop- 2.4. Data on management budgets ulation for which we could find estimates) (sources: African Parks, M. Becker, C. Begg, S. Bhalla, Borana Wildlife Conservancy, A. Cotterill, H. Data on management budgets for protected areas were derived: a) de Longh, K. Everatt, P. Funston, Groom, R., P. Henschel, Lewa Wildlife from grey and published literature (MET, 2010; Cumming, 2012; Conservancy, B. Kissui, R. Kokes, A. Loveridge, G. Maude, A. S., Miller, Tanzania Government, 2013; Packer et al., 2013; Sichilongo et al., Snyman, S. Savini, I. Stevenson, K. Young); and (3) in the absence of re- 2013; Namibia, 2014; Henschel et al., 2014a, 2014b; Nazerali, 2015; sults from recent surveys, from the estimates of the experts interviewed Van Zyl, 2015; Zim Parks, 2015; Games, 2016); b) following direct re- (32.7% of PAs, accounting for 4.3% of the lion population for which we quests for data from wildlife authorities (Botswana, Kenya, Zimbabwe); could find estimates). Expert-based estimates of lion numbers were re- c) from Management Effectiveness and Tracking Tool assessments; c) lied upon only where lion populations are known to be very small or ab- through the surveys of experts affiliated with PAs. For instances where sent. For statistical modelling, we considered PAs to be currently respondents were not clear on the extent of donor support for the PA, ‘effective’ for lion conservation if lions occurred above 50% of estimated we contacted the donors/NGOs directly. Collecting data on manage- K. We made two exceptions: for Akagera National Park in Rwanda and ment budgets was challenging because of the reluctance of some wild- Majete Wildlife Reserve in Malawi, where lions have been reintroduced life authorities to provide data, the perceived sensitivity of the and are increasing in number (but are not yet at 50% of estimated K) fol- information (Hanks and Attwell, 2003), because data on budgets are lowing the establishment of effective management and injection of sig- often not compiled at the PA level (e.g. they are allocated to regions), nificant donor funding (African Parks, 2015). and because documenting all of the donor support, particularly where The effectiveness of PAs for lion prey populations was based on re- there are multiple small donors, is difficult. However, we are confident sponses to questionnaire surveys. Respondents were asked to estimate that our estimates are of the correct order of magnitude and in many the abundance of medium to large ungulates (lion prey species) relative cases significantly more accurate, and that they constitute the most up to the likely carrying capacity of the PA. To reduce scope for error, we to date and accurate compilation of such data in existence at present. converted these estimates to a binary variable, whereby PAs were con- Details and caveats on management budget data are provided in sidered to be ‘effective’ for prey where populations were N50% of esti- Appendix 4. All budget data were converted to 2015 USD. mated K. Such a distinction required that respondents were simply able to indicate whether prey populations were substantially depleted 2.5. Statistical analyses or not, which respondents could comfortably do. In cases where the situation was considered borderline or where estimates looked question- We assessed how PA effectiveness (modelled separately for lions able, we followed up with the respondents and sought secondary and prey) was associated with various characteristics of PAs and com- information sources to qualify those estimates. We found a large degree plexes. These variables included factors that are known to affect lions of congruence between respondents' estimates and indications from ae- and their prey or would logically be expected to influence conservation rial census data (Appendix 2) (in 87.0% of cases respondents' impres- outcomes (Appendix 5). These variables were categorised as: (a) sions matched those from census data from a sample of 99 (56.6%) of ‘Geographic-socioeconomic’, which were considered characteristic of the PAs). In the remaining 13% of PAs where the estimates from surveys the PA's local environment and generally unmodifiable by management and aerial census data did not correspond, the reason for the discrepan- practices; (b) ‘Management variables’, which were considered modifi- cy may be due to changing circumstances on the ground (as the aerial able by PA managers or governments; and, (c) ‘Threat variables’, censuses were conducted 1–10 years ago). We made an exception for which were considered modifiable by management practices but also Gonarezhou National Park in Zimbabwe, which we considered to be ef- strongly affected by the PA's local environment. fective, a PA that since 2008 has had significant elevation in manage- We used logistic regression to investigate hypothesized relation- ment capacity and funding and where prey species have been ships between “effectiveness” of PAs (for lion, and then prey) and the reintroduced and are increasing following the effects of a devastating variables listed in Appendix 5. We first standardized all variables ac- drought in 2002 and subsequent heavy poaching, but are not yet at cording to (Schielzeth, 2010), such that the magnitude of the slope co- 50% of K (Gandiwa et al., 2013). efficients could be compared within and among models. We started The extent of the threat posed to wildlife by each anthropogenic with univariate models of all covariates, and retained all models with pressure was estimated using survey respondents. Respondents were some empirical support (ΔAICc of ≤7) (Burnham and Anderson, 2004). asked to rate 11 potential threats to wildlife in the PA on a 1–5 scale Models were discarded if the candidate variable was correlated at 140 P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149

Fig. 1. Estimated status and trends of lion and prey populations.

|r| ≥ 0.70 with stronger predictors (as determined by AICc). We then models representing each category were ranked by AICc,withAICc built multivariate models with all possible combinations of this variable weight equated to the degree of support for that model being the set, and model-averaged those models (as in (Grueber et al., 2011)) most explanatory model. with a ΔAICc of ≤2. Lastly, we used a linear regression framework to investigate the re- To determine the relative contributions of geographic-socioeconom- lationship between the candidate variables and estimates of the severity ic and management variables in explaining PA effectiveness, the ﬁnal of the most serious threats in PAs in lion range: bushmeat poaching,

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Lions <50% of K Lion populations declining

Fig. 2. The proportion of PAs in the survey in which PAs are not effective at conserving lions (i.e. occurring at b50% of estimated carrying capacity) and the proportion of PAs where lion populations are declining. P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 141 human encroachment for agriculture or settlement, livestock encroach- Table 2 ment, human-wildlife conflict, and poaching for non-meat body parts Standardized beta coefficients of final logistic regression models using management covariates to predict conservation effectiveness of PAs for lions and their prey. Estimates (each modelled individually). Each global model was visually inspected and standard errors with (*) are model-averaged coefficients from models with for homogeneity of model residuals before the building of model ΔAIC b 2; estimates and standard errors without (*) are for a single top model in which subsets (Zuur et al., 2009). Model selection and model averaging were no other model was within 2 ΔAIC. implemented as described above. Lion Prey β β 3. Results * SE* SE Main Use—neither hunting nor tourism −1.3074 0.6512 −1.2302 0.4030 3.1. Status of lions Main Use—hunting −1.1294 0.6997 −0.4478 0.3494 Main Use—tourism −0.0851 0.3327 0.2471 0.2430 Management budget 3.4643 1.2402 2 There are 329 PAs of N1000 km in lion range. Across all PAs for Estimated prey abundance 0.9141 0.2567 which data were available, lions were estimated to occur at ≥50% of PA is fully fenced 0.5071 0.2434 their estimated carrying capacity in just 34.8% of PAs (n = 46, Fig. 1, PA is partially fenced 0.8559 0.2207 Appendix 6). In Botswana, Kenya, Namibia and South Africa, lions occurred at ≥50% of estimated K and were stable or increasing in a high proportion of PAs (Fig. 2). Several countries/regions had no PAs where lions occurred N50% of estimated K and in several, all lion populations c) Threat variables: were considered to be declining. Regionally, East Africa was estimated to have the highest proportion of PAs with lions at ≥50% of K (55.9% of PAs), followed by Southern (30.1%) and Central Africa (20.0%) (though PAs were generally less effective for lions where the perceived data were missing for many Tanzanian PAs). West Africa had no PAs threats from logging of commercially valuable timber, bushmeat with lions N 50% of K. Lions were estimated to occur at b25% of their po- poaching, legal hunting for meat, and human encroachment were tential in 50.7% (n = 76) of PAs. We estimate that if stocked to carrying higher (Table 3). capacity, the 329 state-owned PAs of N1000 km2 in lion range could Of the three categories of variables, management variables were the hold 83,212 lions (2.6–4.2× the current extant population, (Henschel most important determinants of conservation success for lions, with the et al., 2015). Lions typically occurred at higher densities, (averaging top model for management having 92% of AIC weight compared to top 99.3% of K) in fenced reserves than in unfenced ones (37.4%). In 25.8% models for geographic-socioeconomic and threat variables (Table 4). of fenced reserves, lions occurred above our estimates of K for the species, compared to in 9.8% of unfenced PAs. 3.3. Status of lion prey

3.2. Factors determining whether PAs were successful at conserving lions Prey populations were considered to occur at ≥50% of their potential carrying capacity in 44.5% of PAs surveyed (Figs. 1, 3). Prey populations ≥ a) Geographic-socioeconomic variables: were most frequently 50% of potential in Botswana, Namibia, South Af- rica, Tanzania and Kenya (Fig. 3). Prey populations were most common- ly estimated by respondents to be stable or increasing in PAs in Uganda, ≥ PAs with lion populations 50% of estimated carrying capacity Mozambique, South Africa, Botswana and Namibia (Fig. 3). tended to be in countries with lower human infant mortality, higher GDP, in East Africa (notwithstanding a lack of data on lion populations 3.4. Factors determining whether PAs were successful at conserving lion for many Tanzanian PAs), and in PAs where human settlements did prey not occur within the boundaries (Table 1; see Appendix 7 for model selection tables). a) Geographic-socioeconomic variables: b) Management variables: PAs with lion prey populations ≥ 50% of estimated carrying capacity Effectiveness of PAs for lions was associated with higher manage- tended to be in countries with lower human infant mortality and higher ment budgets, where prey was more abundant, and where the PAs GDP, and in PAs where the proportion of land settled by humans was were completely fenced (Table 2). Lions performed best relative to lower (Table 1; see Appendix 7 for model selection tables). their estimated carrying capacity in areas used for photo tourism, b) Management variables: followed by areas with trophy hunting; they did not typically fare well relative to estimated carrying capacity in areas with neither photo tourism nor trophy hunting of use. Prey populations fared poorly in PAs where neither photographic tourism nor trophy hunting was practiced. Prey populations fared better

Table 1 in PAs where there was at least partial fencing present (Table 2). Standardized beta coefficients of final logistic regression models using geographic-socioeconomic covariates to predict conservation effectiveness of PAs for lions and their prey. Table 3 Lion model Prey model Standardized beta coefficients of final logistic regression models using threat covariates to β SE β SE predict conservation effectiveness of PAs for lions and their prey.

Intercept −0.3399 0.2087 Lion Prey Region—South −1.8813 0.4036 β SE β SE Region—Central/West −1.3615 0.7377 Region—East 0.5364 0.4235 Intercept −1.1809 0.2648 −0.3596 0.1740 Infant mortality −0.8116 0.3275 −0.7427 0.2254 Illegal logging score −0.8632 0.3446 GDP 0.9930 0.2759 0.8247 0.2825 Bushmeat poaching score −0.7123 0.2372 −0.5987 0.1888 Human settlement in PA −0.9943 0.2938 Legal hunting score −0.5491 0.2760 Percent of PA settled by humans −0.7302 0.3348 Human encroachment score −0.5331 0.2615 −0.5126 0.1916 142 P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149

Table 4 3.6. Threats to wildlife in general A comparison of top models containing geographic-socioeconomic, management, and threat variables in logistic regression predicting effectiveness of PAs for lions and their Total threat scores for wildlife in PAs were relatively high in Angola prey. K = number of parameters, ΔAICc is difference in AICc between that model and the top model, ω is AIC weight, and LL is log likelihood. (32 ± 3.0), Zambia (23.7 ± 1.9), Malawi (20.4 ± 3.5) and Tanzania (20.8 ± 2.0) and low in South Africa (8.5 ± 1.0), Botswana (10.1 ± Δ ω K AICc AICc LL 1.2), Namibia (12.5 ± 1.6) and Kenya (12.6 ± 3.6). The major threats Lion to wildlife were considered to be: bushmeat poaching (Fig. 5, mean Management 6 123.38 0 0.92 −55.4 − score ± S.E. - 3.0 ± 0.13 across all PAs); poaching of wildlife for non- Geographic-socioeconomic 6 128.21 4.83 0.08 57.81 ﬂ Threat 5 143.05 19.67 0 −66.32 meat body parts (Fig. 5, 2.79 ± 0.12); human-wildlife con ict (Fig. 5, 2.43 ± 0.10), encroachment with livestock (Fig. 5,2.31±0.14)and Prey human encroachment for settlement and agriculture (Fig. 5,2.03± Geographic-socioeconomic 4 180.77 0 1 −86.26 Management 4 195.98 15.2 0 −93.86 0.13). Legal hunting for meat (0.87 ± 0.11), trophy hunting (0.89 ± Threat 3 201.91 21.13 0 −97.88 0.11), mining (1.13 ± 0.11) and disease (1.42 ± 0.10) were considered markedly less severe. Experts' insights into the geographic distribution of threats were congruent with insights from the literature (Appendix 3). c) Threat variables: 3.7. Conditions under which each threat emerged

a) Geographic-socioeconomic: Prey populations did not fare well in PAs where human encroachment and bushmeat hunting were perceived to be more severe (Table 3). Central/West Africa was associated with highest scores for Of the three categories of variables, geographic-socioeconomic bushmeat hunting, livestock encroachment, human-wildlife conflict, variables were the most important determinants of conservation and poaching for non-meat body parts, while East Africa was associated success for prey, with the top model for geographic-socioeconomic with lowest scores for bushmeat hunting and poaching of wildlife for variables having 100% AIC weight compared to top models for manage- non-meat body parts. The presence of human settlement within PAs ment and threat variables (Table 4). was positively associated with bushmeat hunting, human encroachment, and human-wildlife conflict, while the percentage of land occu- pied by humans within PAs was associated with perceived severity of 3.5. Threats to lions livestock encroachment (Table 6; see Appendix 7 for model selection tables). Bushmeat hunting was perceived to be more severe in larger PAs, Respondents most frequently listed the following threats as the in poorer countries, and where livestock densities in the 20 km around most serious for lions in the surveyed PAs: bushmeat poaching/snaring the PA were lower. To the contrary, livestock encroachment was more (26.7% of respondents); human-wildlife conflict (25.5%), encroachment severe in areas with higher densities of livestock in the areas adjacent with livestock (11.4%); human encroachment (6.3%) (Table 5, Fig. 4). to PAs. Human encroachment and poaching for non-meat body parts Poaching of lions for body parts was not identified as a ubiquitous prob- were worse in smaller PA complexes. For all threats except human- lem, but is clearly an emerging issue. Respondents reported evidence of wildlife conflict, geographic-socioeconomic variables were better pre- targeted poaching of lions for body parts most frequently in West Africa dictors of severity than management variables (Table 7). (42.9% of PAs), Mozambique (35.3%), Central Africa (28.6%), Tanzania (22.2%), Zambia (17.4%) and Zimbabwe (10.5%), with no reports of oc- b) Management: currence from the surveys in other countries (though that is not to say that such incidents do not occur – e.g. limited incidences have been re- Lower reported prey abundance was associated with increased se- corded in Botswana and South Africa (Botswana Department of Wildlife verity of bushmeat poaching, human encroachment, and human-wild- and National Parks, unpublished data, K. Marnewick pers. comm.). life conflict; lower management budget was also associated with more

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Prey biomass <50% of estimated K Prey biomass estimated to be declining

Fig. 3. Proportion PAs by country where prey populations are estimated to be b50% of estimated carrying capacity (K) and the proportion of PAs where prey populations are considered to be declining. P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 143 severe bushmeat hunting (Table 8). Land use played a role in the severity of bushmeat poaching, in that problems were more severe when the 0 0 10.0 8.3 0 0 Mining 0 0 0 0 0 1.2% land was used for neither hunting nor tourism or for hunting only. The 0 presence of perimeter fencing was associated with reduced livestock and human encroachment. Livestock encroachment in PAs was more severe with lower management capacity scores. Lastly, there was a pos- 5.6 0 0 0 14.3 0 27.8 0 0 0 Ceremonial killing 0 0 0 0 0 4.5 itive but weak relationship between more severe threats of poaching for 0 non-meat body parts and the presence of legal trophy hunting within those PAs.

c) Threat: 0 0 10.0 23.1 21.4 5.3 0 0 Loss of prey not due to bushmeat 0 11.1 0 0 0 5.1 The presence of individual threats tended to be strongly associated 4.3 with others (Table 9). The severity of bushmeat hunting was positively associated with that of poaching of wildlife for non-meat body parts, illegal logging, and human encroachment. Encroachment with livestock was associated with human-wildlife conflict, human encroachment, tree cutting for charcoal and higher perceived severity of wildlife dis- 35.3 0.0 0.0 0.0 0.0 0.0 0.0 7.1 by product 0.0 11.1 0 0 0 5.7 eases. Human encroachment was associated with higher threat from 14.3 tree cutting for charcoal, livestock encroachment, human-wildlife conflict, illegal logging, and bushmeat poaching. Human-wildlife conflict was associated with higher perceived threats from livestock and 0 0 30.0 25.0 14.3 5.6 33.3 0 Deforestation Poison - lions as a 0 0 0 7.1 3.7 6.4 human encroachment, poaching of wildlife for non-meat body parts, 0 tree cutting for charcoal, and legal meat hunting. Poaching for non- meat body parts was associated with elevated threat from bushmeat poaching, wildlife diseases, legal meat hunting and excessive trophy quotas. 5.9 45.5 0 0 14.3 0 33.3 0 Small population issues 0 0 25.0 14.3 3.7 8.1 0 4. Discussion

4.1. The state of Africa's PAs

A signiﬁcant proportion of PAs in African savannahs currently appear to be underperforming in ecological, and subsequently, in econom-

ic terms (Lindsey et al., 2011; Lindsey et al., 2014; Van Zyl, 2015). destruction outside Populations of both lions and their prey are reported to occur substan- in various African countries. tially below their likely carrying capacity in the majority of PAs in sever- ’ al countries. Furthermore, the situation may be worse than our analyses Disease Isolation/habitat reveal because our sample is biased towards PAs where NGOs are en- gaged (particularly in West and Central Africa). Populations of both 035.39.1 0 0 18.2 0 0 0 08.3 00 8.3 0 14.3 8.3 7.1 0 0.0 0 0 14.3 14.3 lions and their prey are declining in a high proportion of PAs and on- Lion poaching going population declines and range contractions will not be restricted to unprotected lands. top three threats to lions However, some PAs are performing well, and many are far from ‘ 0 0 0 22.2 0 0 33.3 0 0 Trophy hunting 22.2 7.4 3.7 11.1 what could be described as ‘paper parks’ (Bruner et al., 2001). The pat- 42.9 0 4.8 4.3 tern of performance of lions and their prey is more nuanced than depic- tions of regionally positive trends in Southern Africa and negative trends elsewhere (Craigie et al., 2010; Bauer et al., 2015a). For example, 0 0 0 Livestock incursion 28.6 7.1 42.9 21.4 7.1 ed as being in the both lions and their prey appear to be faring relatively well in PAs in Bo- ﬁ tswana, South Africa and Namibia, but also in Kenya (in contrast to the situation in unprotected rangelands, where wildlife populations appear to be declining precipitously in some areas (Ogutu et al., 2015)). Both incursion lion and prey populations are widely regarded as being depressed in many PAs in Central and West Africa, Angola, Ethiopia, Malawi, Mozam- bique, and Zambia, whereas the picture appears more varied in Uganda, Tanzania and Zimbabwe. In Mozambique, wildlife and lion populations (while still depressed and excluding elephants) appear to be recovering in some PAs following the impacts of the civil war and impacts of the Bushmeat HWC Human bushmeat trade thereafter (Lindsey et al., 2013). Drawing conclusions about lion population trends in Tanzania is especially difﬁcult due to the lack of data on lion numbers there (Bauer et al., 2015c). While wildlife in savannah parks in West and Central African PAs are generally under severe pressure, in keeping with forest parks in those regions (Tranquilli et al., 2014), there is room for optimism regarding the pros- Uganda, n = 4 75.0 100 0 Malawi, n = 9Mozambique, n = 18Namibia, n = 94.1 12South Africa, n = 13 100Tanzania, n = 22 27.3 21.4 47.1 29.4 33.3 33.3 27.3 33.3 92.9 0 0 5.9 61.1 11.1 0 22.2 33.3 5.6 5.6 11.1 Angola, n = 4Botswana n = 11Central Africa, n = 9Ethiopia, n = 12 66.7Kenya, 88.0 100 n = 12 8.3 88.9 100 50.0 11.1 10.0 25.0 28.6 25.0 41.7 71.4 0 50.0 25.0 14.3 50.0 10.0 0 42.9 0 10.0 0 20.0 10.0 West Africa, n = 9 92.9 42.9 0 Zambia, n = 27 92.6 29.6 37.0 3.7 Zimbabwe, n = 22Overall, n = 184 57.1 60.2 57.1 4.3 55.9 15.3 0 15.0 13.6 10.2 9.2 8.5

pects for some areas, such as Zakouma National Park in Chad, which has Table 5 The percentage of PAs in which various threats were identi 144 P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 received significant technical and financial support from NGO partners circumstances. Adequate management budgets are particularly im- and donors and which appears to be performing well. portant (Henschel et al., 2016), and low budgets were associated with the emergence of threats. 4.2. Determinants of success African PAs are grossly under-funded in several countries, which un- dermines the ability of wildlife authorities to tackle human pressures Across Africa, variables related to PA management were the most (James et al., 1999; Mansourian and Dudley, 2008; Lindsey et al., important determinants of effectiveness for the conservation of 2016). Resourcing of PAs varies from relatively sufficient in Kenya and lions, in keeping with previous findings (Henschel et al., 2014a, South Africa (and to a lesser extent Botswana and Namibia) to extreme- 2014b; Bauer et al., 2015a; Henschel et al., 2016) and with patterns ly low in countries such as Ethiopia, Central African Republic, Angola, observed for elephants and rhinos (Leader-Williams et al., 1990). Mozambique (James et al., 1999; Mansourian and Dudley, 2008). This suggests that where resources and management capacity are Funding alone is not sufficient, however, and management capacity adequate, lions can be conserved effectively in a wide range of has a significant additive effect on controlling threats.

Fig. 4. Dark shading indicates status as major threat to lions within that protected area. P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 145

Fig. 5. The ﬁve top-ranked threats to wildlife in general in PAs, and the PAs in which each was considered to be a ‘major threat’ (categorised as such when respondent gave a score of 4–5on a1–5 scale when asked how serious the threat is).

The signiﬁcance of fencing for lion conservation is debated (Packer and prey, respectively, and that it is an effective tool for controlling en- et al., 2013; Creel et al., 2013). Our ﬁndings suggest that full and partial croachment by humans and livestock. As human pressures increase and fencing had a strong positive effect on conservation outcomes for lions land neighbouring PAs becomes more fragmented and densely

Table 6 Standardized beta coefficients of final linear regression model using geographic-socioeconomic covariates to predict the emergence of key threats to wildlife in PAs. Estimates and standard errors with (*) are model-averaged coefficients from models with ΔAIC b 2; estimates and standard errors without (*) are for a single top model in which no other model was within 2 ΔAIC.

Bushmeat Livestock Human encroachment Human-wildlife Poaching for encroachment conﬂict non-meat body parts

β* SE* β SE β SE β SE β SE

Region—South 3.3593 0.1528 1.8080 0.1646 2.2596 0.1487 2.4127 0.1371 3.0532 0.1573 Region—Central/West 3.9964 0.3204 3.8346 0.3524 1.8244 0.3350 2.8261 0.3145 3.3394 0.3546 Region—East 2.4639 0.2353 2.9551 0.2517 2.0290 0.2127 2.5860 0.1958 2.2042 0.2251 GDP −0.5859 0.1184 Human settlement in PA 0.2816 0.1102 1.0056 0.1122 0.2683 0.1061 Area of PA (km2) 0.3188 0.1111 Cattle Density 20 km from PA −0.2384 0.1249 0.3820 0.1443 Percent of PA settled by humans 0.4506 0.1192 Area of PA Complex (km2) −0.3854 0.1207 −0.2987 0.1270 146 P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149

Table 7 Large PAs were no better at conserving lion populations relative to A comparison of top models containing geographic-socioeconomic versus management in their estimated carrying capacities than smaller PAs and were associat- linear regression models predicting the emergence of key threats to wildlife. K = number ed with higher perceived pressure from bushmeat poaching. On the of parameters, ΔAICc is difference in AICc between that model and the top model, ω is AIC weight, and LL is log likelihood. other hand, human encroachment and poaching of wildlife for non- meat body parts was perceived as most severe in smaller PA complexes. Δ ω K AICc AICc LL Large PAs are particularly challenging to protect and manage due to Bushmeat scale and cost, and size does not confer immunity to illegal hunting − Geographic-socioeconomic 8 592.2 0 0.88 287.65 (Kiffner et al., 2012; Midlane, 2014; MacKinnon et al., 2015). Notably, Management 6 596.14 3.94 0.12 −291.81 donors appear to be avoiding investing in the largest PAs, which argu- Livestock encroachment ably represents a missed opportunity for making significant gains for − Geographic-socioeconomic 6 626.03 0 1 306.75 lion conservation. Management 6 667.21 41.18 0 −327.34 Human settlement inside PAs had a negative impact on the conser- Human encroachment vation of lions and prey and was strongly associated with bushmeat − Geographic-socioeconomic 6 606.77 0 1 297.12 poaching, livestock and human encroachment, and human-wildlife con- Management 4 654.74 47.97 0 −323.25 flict, in keeping with (Everatt et al., 2014b). High human densities in Human-wildlife conflict lands adjacent to PAs, however, did not have a particularly strong effect − Management 3 581.43 0 0.98 287.64 on conservation outcomes – suggesting that so long as encroachment is Geographic-socioeconomic 5 589 7.57 0.02 −289.32 prevented within PAs, conservation outcomes need not always be Poaching for non-meat greatly impacted by human population growth. Poverty on a national − Geographic-socioeconomic 5 625.01 0 0.56 307.32 level was associated with poor conservation outcomes for lions and Management 5 625.53 0.52 0.44 −307.58 their prey, and was associated with the emergence of key threats, as is the case for elephants (Blanc, 2013). Consequently, if African countries populated, the case for partial or total fencing of some PAs will likely can achieve sustained economic growth and reduced poverty, the pros- grow (Lindsey et al., 2012b). pects for conservation may improve assuming other factors discussed in Lions and their prey tended to fare most poorly relative to estimates this paper are in place, especially adequate PA budgets. of their potential carrying capacity in PAs where there was no economic utilisation of wildlife, and those with no tourism or trophy hunting were more affected by bushmeat poaching, likely due to the reduced pres- 4.3. Threats to lions and their prey ence of guards and PA staff, reduced management budgets within such PAs and potentially due to reduced motivation on the part of wild- We acknowledge that subjective comparisons of the relative severity life authorities to protect such areas if they do not generate income. of threats has potential to be affected by biases as some may be more Similarly, (Bauer et al., 2015a) hypothesized that lion population de- obvious than others. However, insights into the distribution and sever- clines were steepest in reserves with no population monitoring, which ity of the various threats are congruent with knowledge presented in reflects lack of conservation effort. Photographic tourism was associated the literature (Appendix 2). Bushmeat poaching was identified as the with the best conservation outcomes for lions and prey, in keeping with most serious threat for lions and their prey in PAs, reinforcing the find- the findings of (Tranquilli et al., 2014) for wildlife generally. Tourism is ing of Okello and Kiringe (2004),(Kiringe et al., 2007)and(Tranquilli et generally only viable in the areas with the highest densities of wildlife al., 2014). The belief among respondents that bushmeat is the top threat (Lindsey et al., 2006). affirms the suggestion that PAs are more effective at securing wild lands The effect of trophy hunting as the primary use of a PA on the status against human encroachment than against illegal hunting (Bruner et al., of lions and prey could not be determined due to the high standard er- 2001; Geldmann et al., 2013), though encroachment remains a serious rors in our models (Table 2). However, PAs used primarily for hunting threat. Lions are affected both by direct mortalities in snares and also were associated with elevated bushmeat poaching relative to PAs used through reductions in their prey base (Lindsey et al., 2013). Bushmeat primarily for tourism. To a lesser extent, the presence of trophy hunting poaching tends to occur in tandem with legal logging (Poulsen et al., within a PA was associated with increased severity of poaching of wild- 2009) (and as our data suggest, illegal logging), human encroachment life for non-meat body parts (Table 9). The causes of this relationship are and poaching for non-meat body parts. Human encroachment of PAs not clear, but could conceivably be due to there being inadequate anti- represents a severe threat where it occurs, is strongly associated with poaching in some areas used for hunting or conversely due to there the emergence of a wide array of other threats (in keeping with the being greater vigilance and better information on illegal activities findings of (Tranquilli et al., 2014), is likely to be difficult to control where hunters are present (Lindsey et al., 2016). and near impossible to reverse.

Table 8 Standardized beta coefficients of final linear regression model using management covariates to predict the emergence of key threats to wildlife in PAs. Estimates and standard errors with (*) are model-averaged coefficients from models with ΔAIC b 2; estimates and standard errors without (*) are for a single top model in which no other model was within 2 ΔAIC.

Bushmeat Livestock Human Human-wildlife Poaching for encroachment encroachment conﬂict non-meat body parts

β SE β SE β SE β SE β* SE*

Intercept 2.1431 0.1290 2.5100 0.1041 Main use—neither hunting nor tourism 3.6509 0.2212 2.5365 0.2950 2.8818 0.3846 Main use—hunting 3.5127 0.2178 1.8745 0.2682 2.6251 0.9706 Main use—tourism 2.7478 0.1598 2.5380 0.2028 2.9229 0.3607 Estimated prey abundance −0.3897 0.1169 −0.4781 0.1384 −0.3540 0.1044 Management budget −0.3317 0.1155 PA is partially fenced −0.4440 0.1438 −0.3682 0.1384 Total management score −0.3424 0.1661 Trophy hunting present in PA 0.8499 0.4843 P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 147

Table 9 Standardized beta coefficients of final linear regression model using perceived severities of threat to predict the emergence of other key threats to wildlife in PAs. Estimates and standard errors with (*) are model-averaged coefficients from models with ΔAIC b 2; estimates and standard errors without (*) are for a single top model in which no other model was within 2 ΔAIC.

Bushmeat Livestock Human Human-wildlife Poaching for encroachment encroachment conﬂict non-meat body parts

β SE β SE β* SE* β* SE* β SE

Intercept 3.1702 0.1034 2.3717 0.1194 2.1431 0.1044 2.5100 0.0905 2.8379 0.1072 Human encroachment score 0.3829 0.1197 0.3344 0.1500 0.2463 0.1218 Poaching for non-meat score 0.4766 0.1055 0.2412 0.0968 Illegal logging score 0.4317 0.1188 0.3189 0.1377 Human-wildlife conﬂict score 0.4771 0.1334 0.2253 0.1223 Tree cutting for charcoal score 0.3159 0.1466 0.4627 0.1422 0.2493 0.1188 Disease score 0.3467 0.1218 0.2393 0.1088 Bushmeat poaching score 0.3040 0.1217 0.5133 0.1110 Livestock encroachment score 0.3558 0.1237 0.4024 0.1030 Legal hunting score 0.2416 0.0985 0.2505 0.1214 Mining score 0.2490 0.1144 Excessive quota score 0.2619 0.1205

4.4. Necessary steps Promoting the development of appropriate tourism in individual PAs is likely to yield long-term benefits, both by increasing the social If African governments and the global community allocate sufficient and political sustainability of those areas and by helping to reduce financial resources, and given adequate management capacity, lions can threats to wildlife. Management capacity has a significant positive im- be effectively conserved under a wide range of circumstances. There is pact on conservation outcomes, and so fostering capacity-building and an urgent need for elevated funding and more focused conservation ef- long-term partnerships among donor agencies, NGOs, the private sector fort directed towards PAs in many countries (Henschel et al., 2016). Cur- and wildlife authorities is key (Nyirenda and Nkhata, 2013). Such part- rent levels of support from African governments and the donor nerships should then focus on addressing specific threats to wildlife in community are not adequate in many countries (MacKinnon et al., PAs and building long-term sustainable management structures. The 2015). Without such steps, an increasing number of PAs will fail to fulfil inter-linked nature of threats means that steps to address one issue the objectives for which they were gazetted. Similar conclusions were will yield wider collateral benefits. This is particularly true of human en- reached for tiger conservation in Asia (Walston et al., 2010). We also ac- croachment of PAs, and there is a need for governments to resist politi- knowledge the importance of investing in conservation initiatives on cal pressure to allow human settlement or livestock grazing in PAs (e.g. communal and private lands (Ogutu et al., 2015), particularly those ad- http://www.standardmedia.co.ke/article/2000197917/ps-against-bill- jacent to PAs. Such investment can both increase the effective size of PAs allowing-grazing-in-parks, accessed May 2016), so as to limit collateral and also reduce the costs of managing PAs by increasing local support conservation challenges. for conservation. – On a continental level, funding for PAs needs to increase 3 6 fold 5. Conclusions (Lindsey et al., 2016), and in some countries, by a much greater extent fi (James et al., 1999). Inadequate nancial and human resources under- PAs within lion range have potential to host a cumulative population mines the effectiveness of PA management throughout much of the tro- size that is up to four times larger than the current total free-ranging pics (Balmford and Whitten, 2003; Watson et al., 2014), but nowhere lion population. African PAs present the world with the opportunity to more so than Africa (James et al., 1999). There is a strong case for greater effectively secure the future of lions and many other iconic wildlife spe- support from the international community for African PAs. Africa's wild- cies. However, the window for achieving those gains is closing and life is a global resource that confers existence values for people around without urgent and significant investment to improve management the world (Sylven et al., 2012), but one that imposes costs on the people budgets and management capacity, Africa's PA networks will continue that live with it. The illegal trade in wildlife products is being driven by to deteroriate. As human populations increase, there is likely to be polit- growing international demand (Biggs et al., 2013; Williams, 2015). De- ical and social pressure for the conversion of PAs for alternative land veloped nations pledged to assist the conservation efforts of developing uses. This is particularly likely for depleted PAs that confer few econom- nations by providing approximately USD2 billion per year at the Rio ic or social benefits. Indeed, a significant number of African PAs have al- fi Earth Summit in 1992, but have never ful lled those promises (Miller, ready been degazetted, downgraded in status and downsized (http:// 2014). Investing in PAs is a way that developed countries can promote www.padddtracker.org/, accessed April 2016). Africa's vast PA network sustainable economic growth through stimulation of tourism industries. can thus not be taken for granted, nor can it be assumed that wildlife in For the same reason, African governments should consider allocating PAs is safe by virtue of the legally protected status of those areas. larger budgets for PA management to protect the wildlife resources Supplementary data to this article can be found online at http://dx. and ecosystem services on which tourism industries, the African popu- doi.org/10.1016/j.biocon.2017.01.011. lation and economic growth depend (Lindsey et al., 2014). In the few Af- rican countries where PAs are well funded, there are large and valuable tourism industries. The many countries that do not invest adequately in Acknowledgements their PAs, however, risk losing their lions and other wildlife before ever having the chance to benefitfromthem(Lindsey et al., 2016). In Thanks to Kathleen Fitzgerald of African Wildlife Foundation and addition, African countries and people benefitsignificantly from the Dawn Burnham for proofreading this manuscript and providing valu- ecosystem services delivered by PAs (Van Zyl, 2015). Most Africans able insights. are rural and directly dependent on natural resources for daily survival and as much of a quarter of wealth of low income countries is References ‘ ’ derived from natural capital versus just 2% in developed nations African Parks, 2015. African Parks Annual Report 2015. African Parks, Johannesburg, (Fitzgerald, 2015). South Africa. 148 P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149

Atkinson, R., Flint, J., 2001. Accessing hidden and hard-to-reach populations: snowball re- IUCN/UNEP WCMC, 2007. World Database on Protected Areas, Version 2007, WCMC, search strategies. Soc. Res. Update 33, 1–4. Cambridge, UK, 2007. IUCN and UNEP-World Conservation Monitoring. UK, Balmford, A., Whitten, T., 2003. Who should pay for tropical conservation, and how could Cambridge. the costs be met? Oryx 37, 238–250. James, A., Green, M., Payne, J., 1999. A Global Review of Protected Area Budgets and Staff. Bauer, A., Ryskay, H., 2016. Reconnaissance Visit to Alatash – Dinder Lion Conservation World Commission on Protected Areas/World Conservation Press, Cambridge, UK. Unit, Ethiopia – Sudan Border. WildCru, Oxford, UK. Juffe-Bignoli, D., Burgess, N., Bingham, H., Belle, E., de Lima, M., Deguignet, M., Bertzky, B., Bauer, H., Gebresenbet, F., Mathewos, H., 2016. Site Report: Gambella. NABU/Zoo Leipzig Milam, A., Martinez-Lopez, J., Lewis, E., 2014. Protected Planet Report 2014. UNEP- Project, Addis Ababa, Ethiopia. WCMC, Cambridge. Bauer, H., Packer, C., Funston, P.F., Henschel, P., Nowell, K., 2015a. Panthera leo The IUCN Kiffner, C., Stoner, C., Caro, T., 2012. Edge effects and large mammal distributions in a na- Red List of Threatened Species. IUCN. www.iucnredlist.org. tional park. Anim. Conserv. 97–103. Bauer, H., Kamgang, S.A., Kirsten, I., Tumenta, P., Saleh, A., Henschel, P., Sillero-Zubiri, C., Kiringe, J.W., Okello, M.M., Ekajul, S.W., 2007. Managers' perceptions of threats to the 2015b. Large carnivore abundance in the Benoue ecosystem, North Cameroon. Afr. protected areas of Kenya: prioritization for effective management. Oryx 41, 314. J. Ecol. 54, 235–237. Leader-Williams, N., Albon, S., Berry, P., 1990. Illegal exploitation of black rhinoceros and Bauer, H., Chapron, G., Nowell, K., Henschel, P., Funston, P., Hunter, L.T., Macdonald, D.W., elephant populations: patterns of decline, law enforcement and patrol effort in Packer, C., 2015c. Lion (Panthera leo) populations are declining rapidly across Africa, Luangwa Valley, Zambia. J. Appl. Ecol. 1055–1087. except in intensively managed areas. Proc. Natl. Acad. Sci. U. S. A. 112, 14894–14899. Lindsey, P., Balme, G., Booth, V., Midlane, N., 2012a. The significance of African lions for Becker, M., Watson, F., Droge, E., Leigh, K., Carlson, R., Carlson, A., 2012. Estimating past the financial viability of trophy hunting and the maintenance of wild land. PLoS and future male loss in three Zambian lion populations. J. Wildl. Manag. http://dx. One, e29332. doi.org/10.1002/jwmg.446. Lindsey, P., Balme, G., Becker, M., Begg, C., Bento, C., Bocchino, C., Dickman, A., Diggle, R., Bergl, R.A., Oates, J.F., Fotso, R., 2007. Distribution and protected area coverage of endemic Eves, H., Henschel, P., et al., 2013. The bushmeat trade in African savannas: impacts, taxa in West Africa's Biafran forests and highlands. Biol. Conserv. 134, 195–208. drivers, and possible solutions. Biological Conservation 160, 80–96. Biggs, D., Courchamp, F., Martin, R., Possingham, H.P., 2013. Legal trade of Africa's rhino Lindsey, P.A., Balme, G.A., Funston, P.J., Henschel, P., Hunter, L.T., 2016. Life after Cecil: horns. Science 339, 1038–1039 Conservation. channelling global outrage into funding for conservation in Africa. Conservation Let- Blanc, J., 2013. Status of African Elephant Populations and Levels of Illegal Killing and the ters 9, 296–301. Illegal Trade in Ivory. CITES, Geneva, Switzerland. Lindsey, P.A., Nyirenda, V.R., Barnes, J.I., Becker, M.S., McRobb, R., Tambling, C.J., Taylor, Bouché, P., Renaud, P., Lejeune, P., Vermeulen, C., Froment, J., Bangara, A., Fiongai, O., W.A., Watson, F.G., t'Sas-Rolfes, M., 2014. Underperformance of African protected Abdoulaye, A., Abakar, R., Fay, M., 2010. Has the final countdown to wildlife extinc- area networks and the case for new conservation models: insights from Zambia. tion in Northern Central African Republic begun? Afr. J. Ecol. 48, 994–1003. PloS One 9, e94109. Brockington, D., Igoe, J., 2006. Eviction for conservation: a global overview. Conservation Lindsey, P.A., Romañach, S.S., Tambling, C.J., Chartier, K., Groom, R., 2011. Ecological and and Society 4, 424. financial impacts of illegal bushmeat trade in Zimbabwe. Oryx 45, 96. Bruner, A.G., Gullison, R.E., Rice, R.E., da Fonseca, G.A.B., 2001. Effectiveness of parks in Lindsey, P.A., Alexander, R., Frank, L.G., Mathieson, A., Romañach, S.S., 2006. Potential of protecting tropical biodiversity. Science 291, 125–128. trophy hunting to create incentives for wildlife conservation in Africa where alterna- Burnham, K.P., Anderson, D.R., 2004. Multimodel inference understanding AIC and BIC in tive wildlife-based land uses may not be viable. Anim. Conserv. 9, 283–291. model selection. Sociological Methods & Research 33, 261–304. Lindsey, P.A., Masterson, C.L., Beck, A.L., Romañach, S., 2012b. Ecological, social and finan- Canning, D., Raja, S., Yazbeck, A.S., 2015. Africa's Demographic Transition: Dividend or Di- cial issues related to fencing as a conservation tool in Africa. In: Somers, M.J., saster? World Bank Publications Hayward, M. (Eds.), Fencing for Conservation. Springer, New York, pp. 215–234. Caro, T., 2015. Conservation in the African anthropocene. Anonymous Protecting the Lindsey, P.A., Alexander, R., Mills, M.G.L., Romanach, S., Woodroffe, R., 2007. Wildlife Wild. Springer, pp. 164–169. viewing preferences of visitors to protected areas in South Africa: implications for Cozzi, G., Broekhuis, F., McNutt, J., Schmid, B., 2013. Density and habitat use of lions and the role of ecotourism in conservation. J. Ecotourism 6, 19–33. spotted hyenas in northern Botswana and the influence of survey and ecological var- Loveridge, A.J., C., S., 2009. Report on the Lion Distribution and Conservation Modelling iables on call-in survey estimation. Biodivers. Conserv. 22, 2937–2956. Project. Born Free, UK. Craigie, I.D., Baillie, J.E.M., Balmford, A., Carbone, C., Collen, B., Green, R.E., Hutton, J.M., Macdonald, E., Burnham, D., Hinks, A., Dickman, A., Malhi, Y., Macdonald, D., 2015. Con- 2010. Large mammal population declines in Africa's protected areas. Biol. Conserv. servation inequality and the charismatic cat: Felis felicis. Glob. Ecol. Conserv. 3, 143, 2221–2228. 851–866. Creel, S., Becker, M.S., Durant, S.M., M'Soka, J., Matandiko, W., Dickman, A.J., Christianson, MacKinnon, J., Aveling, C., Olivier, R., Vives, M., Saracco, F., 2015. Larger Than Elephants: D., Dröge, E., Mweetwa, T., Pettorelli, N., et al., 2013. Conserving large populations of Inputs for an EU Strategic Approach to Wildlife Conservation in Africa. European lions? The argument for fences has holes. Ecol. Lett. 16 (1413-e3). Commission, Brussels, Belgium. Cumming, D., 2012. A Needs Assessment for Wildlife Protection and Law Enforcement. Mansourian, S., Dudley, N., 2008. Public Fund to Protected Areas. WWF, Gland, Ministry of Environment and Tourism, Windhoek, Namibia. Switzerland. ESRI, 2016. ArcGIS Desktop: Release 10.3.1 Redlands, CA: Environmental Systems Re- MET, 2010. State of Protected Areas in Namibia: A Review of Progress and Challenges. search Institute. www.esri.com. Ministry of Environment and Tourism, Windhoek, Namibia. Everatt, K.T., Andresen, L., Somers, M.J., 2014a. Trophic scaling and occupancy analysis re- Midlane, N., 2014. The Conservation Status and Dynamics of a Protected African Lion veals a lion population limited by top-down anthropogenic pressure in the Limpopo Panthera leo Population in Kafue National Park, Zambia Cape Town, South Africa. National Park. Mozambique, PLoS ONE e99389. Midlane, N., O'Riain, M.J., Balme, G.A., Hunter, L.T., 2015. To track or to call: comparing Everatt, K.T., Andresen, L., Somers, M.J., 2014b. The influence of prey, pastoralism and methods for estimating population abundance of African lions Panthera leo in Kafue poaching on the hierarchical use of habitat by an apex predator. Afr. J. Wild. Res. National Park. Biodivers. Conserv. 1-17. 45, 187–196. Miller, D.C., 2014. Explaining global patterns of international aid for linked biodiversity Ferreira, S.M., Govender, D., Herbst, M., 2013. Conservation implications of Kalahari lion conservation and development. World Dev. 59, 341–359. population dynamics. Afr. J. Ecol. 51, 176–179. Namibia, 2014. Estimates of Revenue, Income and Expenditure 01 April 2014 to 31 March Fitzgerald, K.H., 2015. The silent killer: habitat loss and the role of African protected 2017, Republic of Namibia, Windhoek, Namibia. areas to conserve biodiversity. Anonymous Protecting the Wild. Springer, Nazerali, S., 2015. Sustainable Financing of the System of Protected Areas of Mozambique. pp. 170–188. Verde Azul, Maputo, Mozambique. Games, I., 2016. Hwange National Park General Management Plan 2016–2026. Zimbabwe Nyirenda, V.R., Nkhata, B.A., 2013. Collaborative Governance and Benefit Sharing in Liuwa Parks and Wildlife Management Authority, Harare, Zimbabwe. Plain National Park, Western Zambia. Gandiwa, E., Heitkönig, I.M., Lokhorst, A.M., Prins, H.H., Leeuwis, C., 2013. Illegal hunting Ogutu, J., Piepho, H., Said, M., Ojwang, G., Kifugo, S., Wargute, P., Njino, L., 2015. Extreme and law enforcement during a period of economic decline in Zimbabwe: a case wildlife declines and concurrent increase in livestock numbers in Kenya rangelands. study of northern Gonarezhou National Park and adjacent areas. J. Nat. Conserv. 21, PLoS One 11 (9), e0163249. http://dx.doi.org/10.1371/journal.pone.0163249. 133–142. Okello, M.M., Kiringe, J.W., 2004. Threats to biodiversity and their implications in Geldmann, J., Barnes, M., Coad, L., Craigie, I.D., Hockings, M., Burgess, N.D., 2013. Effective- protected and adjacent dispersal areas of Kenya. Journal of Sustainable Tourism 12, ness of terrestrial protected areas in reducing habitat loss and population declines. 55–69. Biol. Conserv. 161, 230–238. Olléova, M., Dogringar, S., 2013. Zakouma National Park: Carnivore Monitoring Pro- Grueber, C., Nakagawa, S., Laws, R., Jamieson, I., 2011. Multimodel inference in ecology gramme. African Parks, Ndjaména, Chad. and evolution: challenges and solutions. J. Evol. Biol. 24, 699–711. Omoya, E.O., Mudumba, T., Buckland, S.T., Mulondo, P., Plumptre, A.J., 2014. Estimating Hanks, J., Attwell, C., 2003. Financing Africa's Protected Areas, 5th World Parks Congress, population sizes of lions Panthera leo and spotted hyaenas Crocuta crocuta in Durban, South Africa. Uganda's savannah parks, using lure count methods. Oryx 48, 394–401. Henschel, P., Bauer, H., Sogbohoussou, E., Nowell, K., 2015. IUCN Red List: Panthera leo Packer, C., Loveridge, A., Canney, S., Caro, T., Garnett, S., Pfeifer, M., Zander, K., Swanson, A., (West Africa Subpopulation). IUCN, Gland. MacNulty, D., Balme, G., 2013. Conserving large carnivores: dollars and fence. Ecol. Henschel, P., Coad, L., Burton, C., Chataigner, B., Dunn, A., MacDonald, D., Saidu, Y., Hunter, Lett. 16, 635–641. L.T., 2014a. The lion in West Africa is critically endangered. PloS One 9, e83500. Poulsen, J.R., Clark, C.J., Mavah, G., Elkan, P.W., 2009. Bushmeat supply and consumption Henschel, P., Frank, L., Ekwanga, S., Ryan, S., Kuloba, C., Musyoki, C., 2014b. A Survey of in a tropical logging concession in Northern Congo. Conserv. Biol. 23, 1597–1608. Lions and Other Large Carnivore in the Tsavo East and West National Parks. Panthera, Riggio, J., Jacobson, A., Dollar, L., Bauer, H., Becker, M., Dickman, A., Funston, P., Groom, R., New York. Henschel, P., de Iongh, H., 2012. The size of savannah Africa: a lion's (Panthera leo) Henschel, P., Petracca, L.S., Hunter, L., Kiki, M., Sewade, C., Tehou, A., Robinson, H., 2016. view. Biodiversity and Conservation 22, 17–35. Determinants of distribution patterns and management needs in a critically endan- Ripple, W.J., Estes, J.A., Beschta, R.L., Wilmers, C.C., Ritchie, E.G., Hebblewhite, M., Berger, J., gered lion Panthera leo population. Front. Ecol. Evol. http://dx.doi.org/10.3389/fevo. Elmhagen, B., Letnic, M., Nelson, M.P., et al., 2014. Status and ecological effects of the 2016.00110. world's largest carnivores. Science 343, 1241484. P.A. Lindsey et al. / Biological Conservation 209 (2017) 137–149 149

Rosenblatt, E., Becker, M.S., Creel, S., Droge, E., Mweetwa, T., Schuette, P.A., Watson, F., Walston, J., Robinson, J.G., Bennett, E.L., Breitenmoser, U., da Fonseca, G.A., Goodrich, J., Merkle, J., Mwape, H., 2014. Detecting declines of apex carnivores and evaluating Gumal, M., Hunter, L., Johnson, A., Karanth, K.U., 2010. Bringing the tiger back from their causes: an example with Zambian lions. Biol. Conserv. 180, 176–186. the brink—the six percent solution. PLoS Biol 8, e1000485. Schielzeth, H., 2010. Simple means to improve the interpretability of regression coefﬁ- Watson, J.E., Dudley, N., Segan, D.B., Hockings, M., 2014. The performance and potential of cients. Methods in Ecology and Evolution 1, 103–113. protected areas. Nature 515, 67–73. Sichilongo, M., Mulozi, P., Mbewe, B., Machala, C., Pavy, J., 2013. Zambian Wildlife Sector Williams, V., Newton, D., Loveridge, A., Macdonald, D., 2016. Bones of Contention: An As- Policy: Impact Analysis and Recommendations for the Future Policy. Africa Technical sessment of the South African Trade in African Lion Panthera leo Bones and Other Environment and Natural Resources Unit, World Bank, Lusaka, Zambia. Body Parts. TRAFFIC, Cambridge, UK. Smith, R.J., Muir, R.D., Walpole, M.J., Balmford, A., Leader-Williams, N., 2003. Governance Williams, V.L., 2015. Traditional medicines: tiger-bone trade could threaten lions. Nature and the loss of biodiversity. Nature 426, 67–70. 523, 290. Sylven, M., Wildstrand, S., Schepers, F., Birnie, N., Teunissen, T., 2012. Rewilding Europe. WTTC, 2016. Travel & Tourism Economic Impact 2015 Sub-Saharan Africa, World Travel WWF, Nijmegen, Netherlands. and Tourism Council, London, UK. Tanzania Government, 2013. A Performance Audit Report on the Management of Wildlife Yirga, G., Gebresenbet, F., Deckers, J., Bauer, H., 2014. Status of lion (Panthera leo) and in Game Reserves and Game Controlled Areas, United Republic of Tanzania, Arusha, spotted hyena (Crocuta crocuta) in Nechisar National Park, Ethiopia. Momona Ethiop. Tanzania. J. Sci. 6, 127–137. Tranquilli, S., Abedi-Lartey, M., Abernethy, K., Amsini, F., Asamoah, A., Balangtaa, C., Blake, Zim Parks, 2015. Zimbabwe Parks and Wildlife Management Authority Conservation Bud- S., Bouanga, E., Breuer, T., Brncic, T.M., 2014. Protected areas in tropical Africa: get 2015. Zimbabwe Parks and Wildlife Management Authority, Harare, Zimbabwe. assessing threats and conservation activities. PloS One 9, e114154. Zuur, A., Ieno, E., Walker, N., Saveliev, A., Smith, G., 2009. In: Gail, M., Krickeberg, K., Tumenta, P., Kok, J., Van Rijssel, J., Buij, R., Croes, B., Funston, P., De Iongh, H., Udo de Haes, Samet, J.M., Tsiatis, A., Wong, W. (Eds.), Mixed effects models and extensions in ecol- H., 2010. Threat of rapid extermination of the lion (Panthera leo leo) in Waza Nation- ogy with R. Spring Science and Business Media, New York, NY. al Park, Northern Cameroon. Afr. J. Ecol. 48, 888–894. Van Zyl, H., 2015. The Economic Value and Potential of Protected Areas in Ethiopia. Inde- pendent Economic Researchers, Addis Ababa, Ethiopia.